Varela Et Al 1999

7/31/2019 Varela Et Al 1999

1/41999 Macmillan Magazines Ltd

letters to nature

430 NATURE | VOL 397 | 4 FEBRUARY 1999 | www.nature.com

(34%) responded signicantly better to closer stimuli but wereunaffected by the amplitude of the stimulus; 15 (34%) respondedsignicantly better to higher-amplitude stimuli but were unaffectedby the distance to the stimulus; and 11 (25%) responded signi-cantly better to closer stimuli and, independently, to higher-amplitude stimuli. Amplitude is one of many possible cues thathumans use to determine the distance to an auditory event17.Therefore, we suggest that the amplitude-sensitive neuronsdescribed here use this particular cue to code distance. These

neurons will tend to respond to nearby stimuli because they respondbetter to higher-amplitude sounds. However, more than half of theneurons (59%) code distance by means of some other cue orcombination of cues, such that they respond to nearby stimuliindependently of the amplitude. Reverberation of the sound fromthe walls of the room may be important18. Another possible set ofcues for distance involves familiarity with the sound source19.However, the rst neuron tested in monkey 2 was signicantlydependent on distance even though the monkeyhad never heard thestimulus before. Another possible cue is the difference in amplitudebetween the two ears; a very large difference implies a sound sourceclose to one ear. However, the neurons were sensitive to distanceeven when the stimulus was presented on the midline, that is, whenthe amplitude was equal in both ears. Finally, the calculation of

distance near the head maydepend on the highlycomplex distortingeffect of the head and pinnae on the sound spectrum1. This lasteffect would be especially pronounced at such close distances as10 cm. A full analysis of the relative inuence of these different cueswill require further experiments.

The cortical pathways for auditory spatial processing are not wellunderstood. Perhaps distance information is calculated in a differ-ent brain area and then relayed to the trimodal neurons in PMv.Recently, we studied neurons in a portion of parietal area 7b20,intheupper bank of the later sulcus, and found similar trimodal, tactilevisualauditory neurons (M.S.A.G. and C.G.G., manuscript inpreparation). Area 7b projects to PMv21,22, but whether the trimodalregion of 7b projects to the trimodal region of PMv has not yet beendetermined.

Previous experiments showed that multimodal neurons in PMvencode the locations of nearby objects, within about reachingdistance, through touch, vision, and even visual memory1116. Ourresults show that PMv neurons also represent nearby auditoryspace. Because a high proportion of PMv neurons respond duringmovements of the head, mouth, arms and hands, the purpose of thismultimodal map of space may be to guide movements towards andaround the objects that surround the body23,24. M. . . . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . .

Methods

Two adult M. fascicularis were trained to sit in a primate chair; they did not

perform any task. (For details of the experimental procedures, see ref. 16.)

During daily recording sessions, a microdrive was used to lower an electrode

into PMv. Once a neuron was isolated, it was tested for somatosensory, visual

and auditory responses. Somatosensory receptive elds were plotted by

manipulating the joints and stroking the skin. Visual receptive elds were

plotted with objects presented on a wand. Auditory stimuli included tones,

clicks, claps, jingling keys and other sounds. Controlled tests were done using

white noise (20 22,000Hz) presented over Cambridge Soundworks 3-inch

(76.2 mm) speakers mounted in a circular array around the monkey's head at

ear level. The angular position and distance of the speakers to the head was

adjustable. The sound pressure level of the stimuli was measured at the

monkey's head using a Radio Shack sound level meter, repeatedly calibrated

with a 0.25-inch (6.35 mm) Bruel and Kjaer microphone. Neurons were tested

either with the speaker behind the head, or in the dark, so that the monkey

could not see the distance to the sound source. Eye position was not controlled

during the presentation of auditory stimuli. Some PMv neurons are inuenced

by eye position14,16,25. However, the short latency of the auditory response

eliminates the possibility that it was caused by a change in eye position elicited

by the presentation of the stimulus. In addition, there are no reports of

transient bursts of activity in PMv associated with eye movement, whereas

most of the auditory responses in PMv were transient, short-latency bursts

(Fig. 1b).

Received 29 September; accepted 23 November 1998.

1. Blauert, J. Spatial Hearing: The Psychophysics of Human Sound Localization (transl. Allen, J. S.) (MIT

Press, Cambridge, Massachusetts, 1997).

2. Clifton, R. K., Rochat, P., Robin, D. J. & Berthier, N. E. Multimodal perception in the control of infant

reaching. J. Exp. Psychol. Hum. Percept. Perform. 20, 876886 (1994).

3. Coleman, P. D. An analysis of cues to auditory depth perception in free space. Psychol. Bull. 60, 302

315 (1963).

4. Coleman, P. D. Dual role of frequency spectrum in determination of auditory distance. J. Acoust. Soc.Am. 44, 631632 (1968).

5. Edwards, A. A. Accuracy of auditory depth perception. J. Gen. Psychol. 52, 327329 (1955).

6. Gamble, E. A. Intensity as a criterion in estimating the distance of sounds. Psychol. Rev. 16, 416426

(1909).

7. Gardner, M. B. Distance estimation of 08 or apparent 08-oriented speech signals in anechoic space. J.

Acoust. Soc. Am. 45, 4753 (1969).

8. Mershon, D. H. & Bowers, J. N. Absolute and relative cues for the auditory perception of egocentric

distance. Perception 8, 311322 (1979).

9. von Bekesy, G. Experiments in Hearing(McGraw-Hill, New York, 1960).

10. Suga, N. & O'Neill, W. E. Neural axis representing target range in the auditory cortex of the mustache

bat. Science 206, 351353 (1979).

11. Gentilucci, M. et al. Functional organization of inferior area 6 in the macaque monkey. I. Somatotopy

and the control of proximal movements. Exp. Brain. Res. 71, 475490 (1988).

12. Fogassi, L. et al. Coding of peripersonal space in inferior premotor cortex (area F4). J. Neurophysiol.

76, 141157 (1996).

13. Graziano,M. S. A.,Yap, G.S. & Gross, C.G. Coding ofvisual spaceby pre-motorneurons. Science 266,

10541057 (1994).

14. Graziano, M.S. A.,Hu,X.& Gross,C.G. Codingthelocations ofobjectsin thedark. Science 277, 239

241 (1997).

15. Rizzolatti, G. et al. Afferent properties of periarcuate neurons in macaque monkeys. II. Visualresponses. Behav. Brain Res. 2, 147163 (1981).

16. Graziano, M. S. A., Hu, X. & Gross, C. G. Visuo-spatial properties of ventral premotor cortex. J.

Neurophysiol. 77, 22682292 (1997).

17. Ashmead, D. H., LeRoy, D. & Odom, R. D. Perception of the relative distances of nearby sound

sources. Percept. Psychophys. 47, 326331 (1990).

18. Mershon, D. H. & King, L. E. Intensity and reverberation as factors in the auditory perception of

egocentric distance. Percept. Psychophys. 18, 409415 (1975).

19. Coleman, P. D. Failure to localize the source distance of an unfamiliar sound. J. Acoust. Soc. Am. 34,

345346 (1962).

20. Graziano, M. S. A., Fernandez, T. & Gross, C. G. A comparison of bimodal, visual-tactile neurons in

parietal area 7b and ventral premotor cortex of the monkey brain. Neurosci. Abs. 22, 398 (1996).

21. Cavada, C. & Goldman-Rakic, P. S. Posterior parietal cortex in rhesus monkey: II: Evidence for

segregated corticocortical networks linking sensory and limbic areas with the frontal lobe. J. Comp.

Neurol. 287, 422445 (1989).

22. Matelli, M., Camarda, R., Glickstein, M. & Rizzolatti, G. Afferent and efferent projections of the

inferior area 6 in the macaque monkey. J. Comp. Neurol. 255, 281298 (1986).

23. Graziano, M. S. A. & Gross, C. G. Spatial maps for the control of movement. Curr. Opin. Neurobiol. 8,

195201 (1998).

24. Rizzolatti, G., Fadiga, L., Fogassi, L. & Gallese, V. The space around us. Science 277, 190191 1997.

25. Boussaoud, D., Barth, T. M. & Wise, S. P. Effects of gaze on apparent visual responses of frontal cortexneurons. Exp. Brain Res. 93, 423434 (1993).

26. Cohen, J. & P. Cohen, P. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences

(Lawrence Earlbaum Associates, Hillsdale, New Jersey, 1983).

Acknowledgements. We thank E. Olson, X. Hu, S. Alisharan, M. E. Wheeler and V. Gomez for their helpduring the experiment.

Correspondence and requests for materials should be addressed to M.S.A.G. (e-mail: [email protected]).

Perception's shadow: long-

distance synchronization

of human brain activity

Eugenio Rodriguez, Nathalie George,Jean-Philippe Lachaux, Jacques Martinerie,Bernard Renault & Francisco J. Varela

Laboratoire de Neurosciences Cognitives et Imagerie Cerebrale (LENA),

CNRS UPR 640, Hopital de la Salpetriere, 47 Boulevard de l'Hopital,

75651 Paris Cedex 13, France. . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . .

Transient periods of synchronization of oscillating neuronaldischarges in the frequency range 3080 Hz (gamma oscillations)have been proposed to act as an integrative mechanism that maybring a widely distributed set of neurons together into a coherentensemble that underlies a cognitive act14. Results of severalexperiments in animals provide support for this idea (see, for

example, refs 410). In humans, gamma oscillations have been

7/31/2019 Varela Et Al 1999


letters to nature

NATURE | VOL 397 | 4 FEBRUARY 1999 | www.nature.com 431

described both on the scalp1116 (measured by electroencephalo-graphy and magnetoencephalography) and in intracorticalrecordings17, but no direct participation of synchrony in a cogni-tive task has been demonstrated so far. Here we record electricalbrain activity from subjects who are viewing ambiguous visualstimuli (perceived either as faces or as meaningless shapes). Weshow for the rst time, to our knowledge, that only face percep-tion induces a long-distance pattern of synchronization, corre-sponding to the moment of perception itself and to the ensuing

motor response. A period of strong desynchronization marks thetransition between the moment of perception and the motorresponse. We suggest that this desynchronization reects a pro-cess of active uncoupling of the underlying neural ensembles thatis necessary to proceed from one cognitive state to another3.

Ten subjects were shown `Mooney' faces (Fig. 1a, b), which areeasily categorized as faces when presented in upright orientation,but usually seen as meaningless black and white shapes whenpresented upside-down18,19. Subjects were asked to report as quicklyas possible whether they had seen a face or not by pressing on one oftwo different keys. On average, 796 2% of upright presentationswere perceived as faces. Conversely, 766 2% of upside-downpresentations were reported as meaningless. We analysed only thecases of upright presentations that were perceived as faces and

inverted presentations that were perceived as meaningless, referredto here as the `perception' and `no-perception' conditions. Theelectroencephalogram (EEG) was recorded through 30 electrodes,and a precise time frequency analysis was carried out up to 100 Hz.

We rst computed the pseudo WignerVille time frequencytransforms20 of single trials and averaged these transforms over alltrials. This procedure is best adapted to detect the so-called`induced' gamma response, which is triggered by, but not phase-locked to, stimulus onset15 (`phase-locking' is synchronization of

oscillations). We obtained two induced gamma-activity peaks(Fig. 1c, d). The rst induced response, lying at 36 6 3Hz forboth conditions, peaked at ,230ms after stimulus onset, wassignicantly larger for the perception condition (Wilcoxon T 5,Z 2:29, P, 0:05) and has been consistently described as thecorrelate of the perceptual process itself1117. Similar conclusions arereached by studies of evoked potential21. In contrast, the secondinduced gamma peak has not been reported so far. It peaked at,800 ms with a maximum frequency of 406 5 Hz, following thesubject's reaction time closely, and was slightly (but not signi-cantly) stronger in the no-perception than in the perceptioncondition. The latency of this peak indicates the possible involve-ment of post-perceptual processes.

We then studied phase synchrony, the main focus of our work. We

Figure 1 Stimuli and emissiontimefrequencycharts. a, b, Examplesof `Mooney'

faces, high-contrast pictures of a human face. These pictures are easily

recognized as human faces when seen upright ( a), but are difcult to recognize

when inverted (b). c, d, Spectral power following stimulation. A timefrequency

transform was computed in each trial and then summed over all trials, subjects

and electrodes. The chart retains mainly the induced component of the gamma

response. Both charts exhibit two periods of increased gamma-power emission

(between 20 and 60 Hz). Power peaks at ,230ms after stimulus onset, and

between 33 and 39 Hz. The perception condition elicits a signicantly stronger

gamma response than the no-perception condition (Wilcoxon T 5, Z 2:29,

P, 0:05). The second peak lies at ,800ms and 406 5 Hz; it follows after the

reaction time (6456 20 ms for perception; 7666 22 ms for no-perception) and no

signicant differences between conditions are found. The colour scale is indi-

cated in standard deviations, calculated from the 500-ms baseline. We are

summarizing our results here, but 7 out of 10 individual results were also

signicant.

Figure 2 Time courses of phase synchrony and gamma activity. Results shown

are averages over trials, electrodes, and subjects. Values are in standard

deviations fromthe 500-ms baseline.Thickline, face-perception condition(P); thin

line, no-perception condition (NP); dashed line, synchronycomputed on shufed

data (Sh); the grey strip indicates dispersion of these data 63 standard errors;

horizontal redlines indicate6standarderror of reactiontime. a, Phasesynchrony

for the P and NP conditions. NP synchrony remains stable and near the shufing

average until 700 ms. Phase synchrony for the P condition increases at 230 ms

(P, 0:05), decreases sharply at 500ms (P 0:005; absolute value of 0.3), and

endswith a second increase.Synchronyis measuredwith a 250-ms longwavelet;

a commensurable spread of synchrony follows, thusexplainingresultsthat seem

to begin before stimulation onset. Inset, phase-lag distributions for ten increas-

ingly distant synchronic electrode pairs; for all distributions m 0 and j< 408. b,

Gamma-power activity in the 3440 Hz band. In contrast to phase synchrony, and

despite the presence of a signicant difference in synchrony at 250ms, both

conditions follow a similar time course of gamma-band activity.

7/31/2019 Varela Et Al 1999


letters to nature

432 NATURE | VOL 397 | 4 FEBRUARY 1999 | www.nature.com

introduced an effective method22 with which to directly measurephase-locking; this method is based on wavelet ltering followed bytrial-by-trial comparison of phase differences, and overcomes theshortcomings of other techniques used previously5,23,24. Electricalactivity was taken to be synchronous if phase lag between twoelectrodes remained constant throughout all the trials.

The main results obtained from our studies of phase synchronyare summarized in Fig. 2a and compared to gamma activity inFig. 2b. Phase synchrony differed markedly between the perception

and no-perception conditions (Fig. 2a). Only the perception con-dition elicited a three-part temporal pattern spanning the entireduration of the task, from stimulus presentation to motor response.First, a signicant increase in synchrony for the perception relativeto the no-perception condition occurred at ,200260ms afterstimulus presentation (Wilcoxon T 8, Z 1:99, P, 0:05); thetemporal coincidence of this increase in synchrony with the rstgamma response indicates a functional involvement of thisincrease in perception itself. It was followed by a marked decreasein synchrony, or desynchronization, centred at 500 ms(Wilcoxon T 0, Z 2:8, P, 0:005). A nal increase in syn-chrony was present for both conditions at about the reaction time(6456 20 ms of standard error for perception; 7666 22ms fornon-perception). Furthermore, synchronic electrodes had zero-

centred phase-lag distributions, regardless of interelectrode distance(Fig. 2, inset). In contrast to synchrony, gamma activity for boththe perception and the no-perception conditions did not yieldqualitative differences, but only an amplitude difference at 230 ms(Fig. 2b).

The stage of sharp decrease of phase synchrony has not beendescribed before, to our knowledge, but it appears as a verysignicant effect in our results (P 0:005). Given the methodused for computing synchrony, this decrease cannot be interpretedas a return to the baseline level of synchrony (see Methods). Wesuggest that it reects a process of active desynchronization. Thisresult provides the rst direct support for the previous proposal3

that a transition between two distinct cognitive acts (such as faceperception and motor response) should be punctuated by a tran-

sient stage of undoing the preceding synchrony and allowing for theemergence of a new ensemble, through cellular mechanisms thatremain to be established.

To validate our results, we compared them with synchroniescomputed in shuf ed trials, a technique used widely in single-cellstudies4,9,10. Shufing is done by randomizing the order of trials andcalculating synchronies between events that were not recorded atthe same time. This allows an estimation of the magnitude of thebackground random uctuations for the values of synchrony asmeasured here. The average of shufed synchronies varied little overthe entire observation window and showed limited variance, incontrast to both of our experimental conditions (Fig. 2a). Thus ourresults do not represent random phase coincidences present in scalprecordings.

Detailed spatiotemporal information is provided by the regionaldistribution of gamma activity and phase synchrony over thescalp (Fig. 3). The pattern of gamma activity was spatially homo-geneous and was similar between the perception and no-perceptionconditions over time, differing only in amplitude. In contrast, thepattern of synchrony was spatially unhomogeneous and differedover time between conditions. Compared with the no-perceptioncondition, which showed few synchronous patterns, the perceptioncondition exhibited a sequence of localized spatial patterns thatevolved over time (Fig. 3). Synchrony rst increased in the areabetween the left parieto-occipital and frontotemporal regions.Desynchronization was then observed between the parietal andoccipitotemporal areas bilaterally. Parietal regions are involved invisual perception25,26 and episodic memory27. Interactions betweenthese regions and occipitotemporal regions, in particular the fusi-

form gyrus, have been linked to perceptual learning of degraded

Mooney-like faces25. We propose that phase interactions betweenparietal and occipitotemporal regions are essential in the large-scaleintegration that is needed for the perception of upright Mooneys asfaces. The second synchrony increase, which is probably linked tomotor response, was predominant between the right temporal andcentral regions. Only during this second period of increase werethere slight similarities between the perception and no-perceptionconditions, because the subject responded in both cases.

Figure 3 The shadow of a perception. Average scalp distribution of gamma

activity and phase synchrony. Colour coding indicates gamma power (averaged

in a 3440-Hz frequency range) over an electrode and during a 180-ms time

window, from stimulation onset (0 ms) to motor response (720ms). Gamma

activity is spatially homogeneous and similar between conditions over time. In

contrast, phase synchrony is markedly regional and differs between conditions.

Synchrony between electrode pairs is indicated by lines, which are drawn only if

thesynchronyvalueis beyondthedistribution of shufeddata sets (P, 0:01;see

Methods).Blackand green linescorrespondto a signicant increaseor decrease

in synchrony, respectively.

7/31/2019 Varela Et Al 1999


letters to nature

NATURE | VOL 397 | 4 FEBRUARY 1999 | www.nature.com 433

The biological signicance of phase synchrony computed fromscalp EEGs has been doubted because it is difcult to rule out theoccurrence of spurious synchronization resulting from volumeconduction23. However, such spurious synchronization cannotaccount for our results. Volume conduction induces a pattern ofsynchronization that decays rapidly as the separation betweenelectrodes increases beyond 2 cm on the surface of the cortex23. Incontrast, we found that synchrony can be established betweenrecording sites situated far away from each other on the head

surface; only 7% of the synchronies reported here were betweenneighbouring recording sites. It is also conceivable that distantsynchronization could result from a powerful deep source thatdiffuses widely over the scalp. But if this were the case, the phase-synchrony pattern should coincide with gamma activity, the elec-trodes with higher emission being the most synchronous ones. Ourresults do not show this (Fig. 3). Finally, the desynchronizingperiods found here are impossible to explain by volume conduction.If synchrony effects were just a reection of gamma activity thendesynchronization should be associated with periods of low gammaactivity. Our results show the opposite effect: desynchronization co-existed with periods of above-average gamma activity. Furthermore,the same gamma level led to strong desynchronization in theperception condition but had no effect on the no-perception

condition (Fig. 2). Finally, the zero-centred phase-lag distributionfound here also suggests that the measured synchronies didnot havean artefactual origin.

The nding of gamma oscillations has often been taken, erro-neously, as an indication of synchrony. Indeed, changes in gamma-band spectral content cannot be con ated with the phase synchronybetween pairs of electrodes at which such gamma activity occurs.Only synchrony measures bear directly on the possible role ofgamma activity in cognition, as synchrony provides direct informa-tion about electrode pairs and their regional location, which poweremission alone cannot provide. To our knowledge, our results arethe rst to support the theory that phase synchrony is directlyinvolved in human cognition. The long-range character of the phasesynchrony indicates that gamma-phase synchrony (and desyn-

chrony) may be viewed as a mechanism that subserves large-scalecognitive integration2,3,5,8, and not just local visual-feature binding.Finally, we stress that the detection of phase synchrony/desynchronyover the scalp amounts to a dynamic brain mapping that is essentialfor the study of the neural basis of cognitive tasks. M. . . . . . . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . .

Methods

Protocol and recordings. Experimental design19 and data analysis22 have been

described in detail elsewhere. In brief, Mooney faces were presented randomly

for 200 ms in upright or inverted position, in a total of 320 trials. Ten subjects

(2030 years old; seven females) gave their response by pressing buttons

situated under their right and left index ngers. EEGs were recorded by 30

electrodes at standard 1020 positions, to which we added a lower row (M1,

M2, P9, P10, PO9, PO10, O9, and O10). Sampling was taken at 500 Hz.

Timefrequency analysis. After automatic correction of eye-movement artefact

trials, signals were high-pass-ltered at 15 Hz and timefrequency-analysed using

the pseudo WignerVille transformation17,20. Resultingtimefrequencymaps were

normalized and averaged through trials, electrodes and subjects.

Phase-synchrony detection. Previous methods for measuring phase

synchrony between electrode pairs have included spectral coherence5,23,

which mixes energy and phase information, and detection of maximal values

after ltering24, which is inaccurate and slow when large data sets are involved.

In the method introduced here, for each subject phase synchrony was

computed onlyfor thefrequencyf0 of his/hermaximal gamma activity (varying

from 35 to 45 Hz, depending on the subject). Phase was measured from

narrow-band-ltered signals (f0 6 3 Hz) by convolution with a complex

Gabor wavelet designed for f0. An instantaneous phase value, Ji(f0, t, k),

which is a complex number of unit magnitude, was thus obtained for one

electrode, i, a chosen frequency,f0, at time bin indexed byt, and trial k. For each

electrode pair, i and j, and time t, and for all of the k 1;

;N trials, a global

phase-locking value Jij(f0, t) is computed as:

Jijf; t ^kJi2Jj

N

Jij is a real value bounded between 1 (if phase difference is constant) and 0 (if

phase difference is random).

Normalization. To calculate synchrony values comparable between near

(,2 cm) and distant electrode pairs, we carried out a normalization procedure so

that the Jij(f, t) values were compared with the 500-ms baseline preceding the

stimulus. GivenJij,let mij and jij be themean and standard deviation computedfrom a 500-ms prestimulus baseline; the normalized phase-locking values are

then computed as Jij Jij2mij=jij. The same normalization procedure is

applied to timefrequency matrices on a frequency-by-frequency basis.

Topographical synchrony. To display the lines indicating synchrony over

individual pairs of electrodes (Fig. 3) we used the following statistical

procedure. Let Wk be a 180-ms time window between stimulus arrival and

motor response, and let Jij(Wk) be the average phase synchrony between

electrodes i andjover theentire time window, Wk. To enhance time changes, we

compared Wk with the previous time window, Wk-1, and dened the phase-

locking value between pairs nally as DJijWk JijWk2JijWk21. For

each DJij(Wk), 200 values were analogously computed on shufed data

DJsij(Wk). A DJij value is retained as statistically signicant only if greater than

(or lesser than) any of the 200 shufed values DJsij, thus corresponding to a

two-tailed probability value ofP 0:01.Received 7 October; accepted 23 December 1998.

1. Freeman, W. J. Mass Action in the Nervous System (Academic, New York, 1975).

2. Damasio, A. R. Synchronous activation in multiple cortical regions: a mechanism for recall. Semin.

Neurosci. 2, 287297 (1990).

3. Varela, F. J. Resonant cell assemblies: a new approach to cognitive function and neuronal synchrony.

Biol. Res. 28, 8195 (1995).

4. Singer, W. & Gray, C. M. Visual feature integration and the temporal correlation hypothesis. Annu.

Rev. Neurosci. 18, 555586 (1995).

5. Bressler, S. L., Coppola, R. & Nakamura, R. Episodic multiregional cortical coherence at multiple

frequencies during visual task performance. Nature 366, 153156 (1993).

6. Gray, C.M.,Konig,P.,Engel,A. K. & Singer,W.Oscillatoryresponsesin catvisual cortex exhibitinter-

columnar synchronization which reects global stimulus properties. Nature 338, 334337 (1989).

7. Vaadia, E. et al. Dynamics of neuronal interaction in the monkey cortex in relation to behavioral

events. Nature 373, 515518 (1995).

8. Roelfsema, P. R., Engel, A. K., Konig, P. & Singer, W. Visuomotor integration is associated with the

zero time-lag synchronization among cortical areas. Nature 385, 157161 (1997).

9. Neuenschwander, S., Engel, A., Konig, P., Singer, W. & Varela, F. Synchronization of neuronal

responses in the optic tectum of awake pigeons. Vis. Neurosci. 13, 575584 (1996).

10. Kreiter, A. K. & Singer, W. Stimulus-dependent synchronization of neuronal responses in the visual

cortex of the awake macaque monkey. J. Neurosci. 16, 23812396 (1996).

11. Llinas, R. & Ribary, U. Coherent 40Hz oscillation characterizes dream state in humans. Proc. Natl

Acad. Sci. USA 90, 20782081 (1993).

12. Tiitinen, H. et al. Selective attention enhances the auditory 40-Hz transient response in humans.

Nature 364, 5960 (1993).

13. Desmedt, J. E. & Tomberg, C. Transient phase-locking of 40 Hz electrical oscillations in prefrontal

parietal cortex reects the process of conscious somatic perception. Neurosci. Lett. 168, 126129

(1994).

14. Joliot, M., Ribary, U. & Llinas, R. Human oscillatory brain activity near 40 Hz coexists with cognitive

temporal binding. Proc. Natl Acad. Sci. USA 91, 1174811751 (1994).

15. Pantev, C. Evoked and induced gamma band activity of the human cortex. Brain Topogr. 7, 321330

(1995).

16. Tallon-Baudry, C., Bertrand, O., Delpuech, C. & Pernier, J. Oscillatory gamma band (30 70 Hz)

activity induced by a visual search task in human. J. Neurosci. 17, 722734 (1997).

17. Lachaux, J. P. et al. Gamma band activity in human intracortical recordings triggered by cognitive

tasks. Eur. J. Neurosci. (submitted).

18. Mooney, C. M. Age in the development of closure ability in children. Can. J. Psychol. 11, 219226

(1957).

19. George, N., Jemel, B., Fiori, N. & Renault, B. Face and shape repetition effects in humans: a spatio-temporal ERP study. NeuroReport 8, 14171423 (1997).

20. Mecklenbrauker,W. F.G. (ed.) The Wigner DistributionTheory and Applicationsin Signal Processing

(Elsevier, Amsterdam, 1993).

21. Jeffreys, D. A. Evoked potentials studies of face and object processing. Vis. Cogn. 3, 138 (1996).

22. Lachaux, J. P., Rodriguez, E., Muller-Gerking, J., Martinerie, J. & Varela, F. J. Measuring phase-

synchrony in brain signals. Hum. Brain Mapp. (submitted).

23. Menon, V. et al. Spatio-temporal correlations in human gamma band electrocorticograms. Electro-

encephalogr. Clin. Neurophysiol. 98, 89102 (1996).

24. Yordanova, J., Kolev, V. & Demiralp, T. The phase-locking of auditory gamma band responses in

humans is sensitive to task processing. NeuroReport 8, 39994004 (1997).

25. Dolan,R. J. etal. How thebrain learnstosee objects andfaces inan impoverished context.Nature389,

596599 (1997).

26. Friedman-Hill, S. R., Robertson, L. C. & Treisman, A. Parietal contributionsto visual feature binding:

evidence from a patient with bilateral lesions. Science 269, 853855 (1995).

27. Shallice, T. et al. Brain regions associated with acquisition and retrieval of verbal episodic memory.

Nature 368, 633635 (1994).

Acknowledgements. We thank Y. Okada for suggestions on the manuscript. This work was supported bygrants from MIDEPLAN (Chile), DRET (France), and the Human Science Frontier.

Correspondence and requests for materials should be addressed to F.J.V. (e-mail: [email protected]).

Documents

Varela Et Al 1999